26 Overview of Stata estimation commands

Contents
26.1 Introduction
26.2 Means, proportions, and related statistics
26.3 Linear regression with simple error structures
26.4 Structural equation modeling (SEM)
26.5 ANOVA, ANCOVA, MANOVA, and MANCOVA
26.6 Generalized linear models
26.7 Binary-outcome qualitative dependent-variable models
26.8 ROC analysis
26.9 Conditional logistic regression
26.10 Multiple-outcome qualitative dependent-variable models
26.11 Count dependent-variable models
26.12 Exact estimators
26.13 Linear regression with heteroskedastic errors
26.14 Stochastic frontier models
26.15 Regression with systems of equations
26.16 Models with endogenous sample selection
26.17 Models with time-series data
26.18 Panel-data models
26.18.1 Linear regression with panel data
26.18.2 Censored linear regression with panel data
26.18.3 Generalized linear models with panel data
26.18.4 Qualitative dependent-variable models with panel data
26.18.5 Count dependent-variable models with panel data
26.18.6 Random-coefficients model with panel data
26.19 Multilevel mixed-effects models
26.20 Survival-time (failure-time) models
26.21 Treatment-effect models
26.22 Generalized method of moments (GMM)
26.23 Estimation with correlated errors
26.24 Survey data
26.25 Multiple imputation
26.26 Multivariate and cluster analysis
26.27 Pharmacokinetic data
26.28 Specification search tools
26.29 Power and sample-size analysis
26.30 Obtaining new estimation commands
26.31 References

26.1 Introduction
Estimation commands fit models such as linear regression and probit. Stata has many such
commands, so it is easy to overlook a few. Some of these commands differ greatly from each other,
others are gentle variations on a theme, and still others are equivalent to each other.

1
 2 [ U ] 26 Overview of Stata estimation commands

Estimation commands share features that this chapter will not discuss; see [U] 20 Estimation and
postestimation commands. Especially see [U] 20.21 Obtaining robust variance estimates, which
discusses an alternative calculation for the estimated variance matrix (and hence standard errors) that
many of Stata’s estimation commands provide, and [U] 20.12 Performing hypothesis tests on the
coefficients.
Here, however, this chapter will put aside all of that — and all issues of syntax — and deal solely
with matching commands to their statistical concepts. This chapter will not cross-reference specific
commands. To find the details on a particular command, look up its name in the index.

26.2 Means, proportions, and related statistics

This group of estimation commands computes summary statistics rather than fitting regression
models. However, being estimation commands, they share the features discussed in [U] 20 Estimation
and postestimation commands, such as allowing the use of postestimation commands.
mean, proportion, ratio, and total provide estimates of population means, proportions, ratios,
and totals, respectively. Each of these commands allows for obtaining separate estimates within
subpopulations, groups defined by a separate categorical variable. In addition, mean, proportion,
and ratio can report statistics adjusted by direct standardization.
pwmean provides another option for computing means of one variable for each level of one or
more categorical variables. In addition, pwmean computes all pairwise differences in these means
along with the corresponding tests and confidence intervals, which can optionally be adjusted for to
account for multiple comparisons.

26.3 Linear regression with simple error structures

Consider models of the form
y j = x j β + j

for a continuous y variable. In this category, estimation is restricted to when σ2 is constant across
observations j . The model is called the linear regression model, and the estimator is often called the
(ordinary) least-squares (OLS) estimator.
regress is Stata’s linear regression command. (regress produces the robust estimate of variance
as well as the conventional estimate, and regress has a collection of commands that can be run
after it to explore the nature of the fit.)
Also, the following commands will do linear regressions, as does regress, but offer special
features:
1. ivregress fits models in which some of the regressors are endogenous, using either instrumental
variables or generalized method of moments (GMM) estimators.
2. areg fits models yj = xj β +dj γ +j , where dj is a mutually exclusive and exhaustive dummy
variable set. areg obtains estimates of β (and associated statistics) without ever forming dj ,
meaning that it also does not report the estimated γ. If your interest is in fitting fixed-effects
models, Stata has a better command—xtreg—discussed in [U] 26.18.1 Linear regression with
panel data. Most users who find areg appealing will probably want to use xtreg because
it provides more useful summary and test statistics. areg duplicates the output that regress
would produce if you were to generate all the dummy variables. This means, for instance, that
the reported R2 includes the effect of γ.
 [ U ] 26 Overview of Stata estimation commands 3

3. boxcox obtains maximum likelihood estimates of the coefficients and the Box – Cox transform
parameters in a model of the form

(θ) (λ) (λ) (λ)

yi = β0 + β1 xi1 + β2 xi2 + · · · + βk xik + γ1 zi1 + γ2 zi2 + · · · + γl zil + i

where  ∼ N (0, σ 2 ). Here the y is subject to a Box–Cox transform with parameter θ. Each of
the x1 , x2 , . . . , xk is transformed by a Box–Cox transform with parameter λ. The z1 , z2 , . . . , zl
are independent variables that are not transformed. In addition to the general form specified
above, boxcox can fit three other versions of this model defined by the restrictions λ = θ,
λ = 1, and θ = 1.
4. tobit allows estimation of linear regression models when yi has been subject to left-censoring,
right-censoring, or both. Say that yi is not observed if yi < 1,000, but for those observations,
it is known that yi < 1,000. tobit fits such models.
ivtobit does the same but allows for endogenous regressors.
5. intreg (interval regression) is a generalization of tobit. In addition to allowing open-ended
intervals, intreg allows closed intervals. Rather than observing yj , it is assumed that y0j and
y1j are observed, where y0j ≤ yj ≤ y1j . Survey data might report that a subject’s monthly
income was in the range $1,500–$2,500. intreg allows such data to be used to fit a regression
model. intreg allows y0j = y1j and so can reproduce results reported by regress. intreg
allows y0j to be −∞ and y1j to be +∞ and so can reproduce results reported by tobit.
6. truncreg fits the regression model when the sample is drawn from a restricted part of the
population and so is similar to tobit, except that here the independent variables are not
observed. Under the normality assumption for the whole population, the error terms in the
truncated regression model have a truncated-normal distribution.
7. cnsreg allows you to place linear constraints on the coefficients.
8. eivreg adjusts estimates for errors in variables.
9. nl provides the nonlinear least-squares estimator of yj = f (xj , β) + j .
10. rreg fits robust regression models, which are not to be confused with regression with robust
standard errors. Robust standard errors are discussed in [U] 20.21 Obtaining robust vari-
ance estimates. Robust regression concerns point estimates more than standard errors, and it
implements a data-dependent method for downweighting outliers.
11. qreg produces quantile regression estimates, a variation that is not linear regression at all but
is an estimator of yj = xj β + j . In the basic form of this model, sometimes called median
regression, xj β measures not the predicted mean of yj conditional on xj , but its median. As
such, qreg is of most interest when j does not have constant variance. qreg allows you to
specify the quantile, so you can produce linear estimates for the predicted 1st, 2nd, . . . , 99th
percentile.
Another command, bsqreg, is identical to qreg but presents bootstrap standard errors.
The sqreg command estimates multiple quantiles simultaneously; standard errors are obtained
via the bootstrap.
The iqreg command estimates the difference between two quantiles; standard errors are obtained
via the bootstrap.
12. vwls (variance-weighted least squares) produces estimates of yj = xj β +j , where the variance
of j is calculated from group data or is known a priori. vwls is therefore of most interest to
categorical-data analysts and physical scientists.
 4 [ U ] 26 Overview of Stata estimation commands

26.4 Structural equation modeling (SEM)

SEM stands for structural equation modeling. The sem and gsem commands fit SEM.
sem fits standard linear SEMs. gsem fits what we call generalized SEMs, generalized to allow for
generalized linear responses and multilevel modeling.
Generalized linear means binary responses such as probit and logit, count responses such as Poisson
and negative binomial, categorical responses such as multinomial logit, and ordered responses such
as ordered probit, ordered logit, and more. Generalized linear includes linear responses.
Multilevel modeling allows for nested effects such as patient within doctor and patients within
doctor within hospital and crossed effects such as occupation and industry.
Let’s start with sem. sem can fit models ranging from linear regression to measurement models to
simultaneous equations, including confirmatory factor analysis (CFA) models, correlated uniqueness
models, latent growth models, and multiple indicators and multiple causes (MIMIC) models. You can
obtain standardized or unstandardized results, direct and indirect effects, goodness-of-fit statistics,
modification indices, scores tests, Wald tests, linear and nonlinear tests of estimated parameters, and
linear and nonlinear combinations of estimated parameters with confidence intervals. You can perform
estimation across groups with easy model specification and easy-to-use tests for group invariance.
All this may be done using raw or summary statistics data. In addition, sem optionally can use full
information maximum-likelihood (FIML) estimation to handle observations containing missing values.
gsem extends the types of models that can be fit. Responses may be continuous, ordinal, count, or
categorical, and gsem allows for multilevel modeling. Latent variables can be included at any level.
This allows for fitting models with random intercepts and random slopes. These random effects may
be nested or crossed.
There is considerable overlap in the capabilities of sem and gsem. Whenever there is overlap, sem
is faster and sometimes easier to use.
The generalized response variables allowed by gsem permit fitting measurement models with
different types of responses, latent growth models with different types of responses, and so on.
gsem can also fit item–response theory (IRT) models, multilevel CFA models, multilevel mixed-effects
models, and multilevel structural equation models.
Where appropriate, results can be reported in exponentiated form to provide odds ratios, incidence-
rate ratios, and relative-risk ratios. You can also obtain predictions, likelihood-ratio tests, Wald tests,
predictive margins, contrasts, and pairwise comparisons.
Whether fitting a model with sem or gsem, you can specify your model by typing the command
or by using the SEM Builder to draw path diagrams.
For those of you unfamiliar with SEM, it is worth your time to learn about it if you ever fit linear
regressions, logistic regressions, ordered logit regressions, ordered probit regressions, Poisson regres-
sions, seemingly unrelated regressions, multivariate regressions, simultaneous systems, measurement
error models, selection models, endogenous treatment-effects models, and multilevel mixed-effects
models, or if you are interested in generalized method of moments.
sem and gsem can be used to fit many models that can be fit by other Stata commands. The
advantage of using sem and gsem is in the extensions that they can provide. They allow for introduction
of latent variables to account for measurement error, simultaneous equations with different types of
responses, multilevel versions of popular models such as selection models, and more.
See the [SEM] Stata Structural Equation Modeling Reference Manual; in particular, see [SEM] intro 5.
 [ U ] 26 Overview of Stata estimation commands 5

26.5 ANOVA, ANCOVA, MANOVA, and MANCOVA

ANOVA and ANCOVA are related to linear regression, but we classify them separately. The related
Stata commands are anova, oneway, and loneway. The manova command provides MANOVA and
MANCOVA (multivariate ANOVA and ANCOVA).
anova fits ANOVA and ANCOVA models, one-way and up—including two-way factorial, three-way
factorial, etc.—and fits nested and mixed-design models and repeated-measures models.
oneway fits one-way ANOVA models. It is quicker at producing estimates than anova, although
anova is so fast that this probably does not matter. The important difference is that oneway can
report multiple-comparison tests.
loneway is an alternative to oneway. The results are numerically the same, but loneway can deal
with more levels (limited only by dataset size; oneway is limited to 376 levels and anova to 798, but
for anova to reach 798 requires a lot of memory), and loneway reports some additional statistics,
such as the intraclass correlation.
manova fits MANOVA and MANCOVA models, one-way and up—including two-way factorial,
three-way factorial, etc.—and it fits nested and mixed-design models.

26.6 Generalized linear models

The generalized linear model is

g{E(yj )} = xj β, yj ∼ F

where g() is called the link function and F is a member of the exponential family, both of which
you specify before estimation. glm fits this model.
The GLM framework encompasses a surprising array of models known by other names, including
linear regression, Poisson regression, exponential regression, and others. Stata provides dedicated
estimation commands for many of these. Stata has, for instance, regress for linear regression,
poisson for Poisson regression, and streg for exponential regression, and that is not all of the
overlap.
glm by default uses maximum likelihood estimation and alternatively estimates via iterated
reweighted least squares (IRLS) when the irls option is specified. For each family, F , there is
a corresponding link function, g(), called the canonical link, for which IRLS estimation produces
results identical to maximum likelihood estimation. You can, however, match families and link func-
tions as you wish, and, when you match a family to a link function other than the canonical link,
you obtain a different but valid estimator of the standard errors of the regression coefficients. The
estimator you obtain is asymptotically equivalent to the maximum likelihood estimator, which, in
small samples, produces slightly different results.
For example, the canonical link for the binomial family is logit. glm, irls with that combination
produces results identical to the maximum-likelihood logit (and logistic) command. The binomial
family with the probit link produces the probit model, but probit is not the canonical link here. Hence,
glm, irls produces standard error estimates that differ slightly from those produced by Stata’s
maximum-likelihood probit command.
Many researchers feel that the maximum-likelihood standard errors are preferable to IRLS estimates
(when they are not identical), but they would have a difficult time justifying that feeling. Maximum
likelihood probit is an estimator with (solely) asymptotic properties; glm, irls with the binomial
family and probit link is an estimator with (solely) asymptotic properties, and in finite samples, the
standard errors differ a little.
 6 [ U ] 26 Overview of Stata estimation commands

Still, we recommend that you use Stata’s dedicated estimators whenever possible. IRLS — the
theory — and glm, irls — the command — are all encompassing in their generality, meaning that
they rarely use the right jargon or provide things in the way you wish they would. The narrower
commands, such as logit, probit, and poisson, focus on the issue at hand and are invariably
more convenient.
glm is useful when you want to match a family to a link function that is not provided elsewhere.
glm also offers several estimators of the variance–covariance matrix that are consistent, even when
the errors are heteroskedastic or autocorrelated. Another advantage of a glm version of a model
over a model-specific version is that many of these VCE estimators are available only for the glm
implementation. You can also obtain the ML–based estimates of the VCE from glm.

26.7 Binary-outcome qualitative dependent-variable models

There are many ways to write these models, one of which is
Pr(yj 6= 0) = F (xj β)
where F is some cumulative distribution. Two popular choices for F () are the normal and logistic, and
the models are called the probit and logit (or logistic regression) models. A third is the complementary
log–log function; maximum likelihood estimates are obtained by Stata’s cloglog command.
The two parent commands for the maximum likelihood estimator of probit and logit are probit
and logit, although logit has a sibling, logistic, that provides the same estimates but displays
results in a slightly different way. There is also an exact logistic estimator; see [U] 26.12 Exact
estimators.
Do not read anything into the names logit and logistic. Logit and logistic have two interchanged
definitions in two scientific camps. In the medical sciences, logit means the minimum χ2 estimator,
and logistic means maximum likelihood. In the social sciences, it is the other way around. From our
experience, it appears that neither reads the other’s literature, because both talk (and write books)
asserting that logit means one thing and logistic the other. Our solution is to provide both logit
and logistic, which do the same thing, so that each camp can latch on to the maximum likelihood
command under the name each expects.
There are two slight differences between logit and logistic. logit reports estimates in the
coefficient metric, whereas logistic reports exponentiated coefficients — odds ratios. This is in
accordance with the expectations of each camp and makes no substantial difference. The other
difference is that logistic has a family of postlogistic commands that you can run to explore the
nature of the fit. Actually, that is not exactly true because all the commands for use after logistic
can also be used after logit.
If you have not already selected logit or logistic as your favorite, we recommend that you
try logistic. Logistic regression (logit) models are more easily interpreted in the odds-ratio metric.
In addition to logit and logistic, Stata provides glogit, blogit, and binreg commands.
blogit is the maximum likelihood estimator (the same as logit or logistic) but applied on
data organized in a different way. Rather than having individual observations, your data are organized
so that each observation records the number of observed successes and failures.
glogit is the weighted-regression, grouped-data estimator.
binreg can be used to model either individual-level or grouped data in an application of the
generalized linear model. The family is assumed to be binomial, and each link provides a distinct
parameter interpretation. Also, binreg offers several options for setting the link function according
to the desired biostatistical interpretation. The available links and interpretation options are
 [ U ] 26 Overview of Stata estimation commands 7

Option Implied link Parameter

or logit Odds ratios = exp(β )
rr log Risk ratios = exp(β )
hr log complement Health ratios = exp(β )
rd identity Risk differences = β

Related to logit, the skewed logit estimator scobit adds a power to the logit link function and is
estimated by Stata’s scobit command.
Turning to probit, you have two choices: probit and ivprobit. probit fits a maximum-likelihood
probit model. ivprobit fits a probit model where one or more of the regressors are endogenously
determined.
Stata also provides bprobit and gprobit. The bprobit command is a maximum likelihood
estimator — equivalent to probit — but works with data organized in the different way outlined above.
gprobit is the weighted-regression, grouped-data estimator.
Continuing with probit: hetprobit fits heteroskedastic probit models. In these models, the variance
of the error term is parameterized.
heckprobit fits probit models with sample selection.
Also, Stata’s biprobit command fits bivariate probit models, meaning two correlated outcomes.
biprobit also fits partial-observability models in which only the outcomes (0, 0) and (1, 1) are
observed.

26.8 ROC analysis

ROC stands for receiver operating characteristics. ROC deals with specificity and sensitivity, the
number of false positives and undetected true positives of a diagnostic test. The term ROC dates back
to the early days of radar when there was a knob on the radar receiver labeled “ROC”. If you turned the
knob one way, the receiver became more sensitive, which meant it was more likely to show airplanes
that really were there and, simultaneously, more likely to show returns where there were no airplanes
(false positives). If you turned the knob the other way, you suppressed many of the false positives,
but unfortunately, also suppressed the weak returns from real airplanes (undetected positives). These
days, in the statistical applications we imagine, one does not turn a knob but instead chooses a value
of the diagnostic test, above which is declared to be a positive, and below which, a negative.
ROC analysis is applied to binary outcomes such as those appropriate for probit or logistic regression.
After fitting a model, one can obtain predicted probabilities of a positive outcome. One chooses a
value, above which the predicted probability is declared a positive and below which, a negative.
ROC analysis is about modeling the tradeoff of sensitivity and specificity as the threshold value is
chosen.
Stata’s suite for ROC analysis consists of six commands: roctab, roccomp, rocfit, rocgold,
rocreg, and rocregplot.
roctab provides nonparametric estimation of the ROC curve and produces Bamber and Hanley
confidence intervals for the area under the curve.
roccomp provides tests of equality of ROC areas. It can estimate nonparametric and parametric
binormal ROC curves.
rocfit fits maximum likelihood models for a single classifier, an indicator of the latent binormal
variable for the true status.
 8 [ U ] 26 Overview of Stata estimation commands

rocgold performs tests of equality of ROC area against a “gold standard” ROC curve and can
adjust significance levels for multiple tests across classifiers via Šidák’s method.
rocreg performs ROC regression; it can adjust both sensitivity and specificity for prognostic factors
such as age and gender; it is by far the most general of all the ROC commands.
rocregplot graphs ROC curves as modeled by rocreg. ROC curves may be drawn across covariate
values, across classifiers, and across both.
See [R] roc.

26.9 Conditional logistic regression

clogit is Stata’s conditional logistic regression estimator. In this model, observations are assumed
to be partitioned into groups, and a predetermined number of events occur in each group. The model
measures the risk of the event according to the observation’s covariates, xj . The model is used in
matched case – control studies (clogit allows 1 : 1, 1 : k , and m : k matching) and is used in natural
experiments whenever observations can be grouped into pools in which a fixed number of events
occur. clogit is also used to fix logistic regression with fixed group effects.

26.10 Multiple-outcome qualitative dependent-variable models

For more than two outcomes, Stata provides ordered logit, ordered probit, rank-ordered logit,
multinomial logistic regression, multinomial probit regression, McFadden’s choice model (conditional
fixed-effects logistic regression), and nested logistic regression.
oprobit and ologit provide maximum-likelihood ordered probit and logit. These are general-
izations of probit and logit models known as the proportional odds model and are used when the
outcomes have a natural ordering from low to high. The idea is that there is an unmeasured zj = xj β,
and the probability that the k th outcome is observed is Pr(ck−1 < zj < ck ), where c0 = −∞,
ck = +∞, and c1 , . . . , ck−1 along with β are estimated from the data.
heckoprobit fits ordered probit models with sample selection.
rologit fits the rank-ordered logit model for rankings. This model is also known as the Plackett–
Luce model, the exploded logit model, and choice-based conjoint analysis.
asroprobit fits the probit model for rankings, a more flexible estimator than rologit because
asroprobit allows covariances among the rankings. asroprobit is also similar to asmprobit
(below), which is used for outcomes that have no natural ordering. The as in the name signifies that
asroprobit also allows alternative-specific regressors—variables that have different coefficients for
each alternative.
slogit fits the stereotype logit model for data that are not truly ordered, as data are for ologit,
but for which you are not sure that it is unordered, in which case mlogit would be appropriate.
mlogit fits maximum-likelihood multinomial logistic models, also known as polytomous logistic
regression. It is intended for use when the outcomes have no natural ordering and you know only
the characteristics of the outcome chosen (and, perhaps, the chooser).
asclogit fits McFadden’s choice model, also known as conditional logistic regression. In the
context denoted by the name McFadden’s choice model, the model is used when the outcomes have
no natural ordering, just as multinomial logistic regression, but the characteristics of the outcomes
chosen and not chosen are known (along with, perhaps, the characteristics of the chooser).
 [ U ] 26 Overview of Stata estimation commands 9

In the context denoted by the name conditional logistic regression — mentioned above — subjects
are members of pools, and one or more are chosen, typically to be infected by some disease or to
have some other unfortunate event befall them. Thus the characteristics of the chosen and not chosen
are known, and the issue of the characteristics of the chooser never arises. Either way, it is the same
model.
In their choice-model interpretations, mlogit and clogit assume that the odds ratios are inde-
pendent of other alternatives, known as the independence of irrelevant alternatives (IIA) assumption.
This assumption is often rejected by the data and the nested logit model relaxes this assumption.
nlogit is also popular for fitting the random utility choice model.
asmprobit is for use with outcomes that have no natural ordering and with regressors that are
alternative specific. It is weakly related to mlogit. Unlike mlogit, asmprobit does not assume the
IIA.
mprobit is also for use with outcomes that have no natural ordering but with models that do not
have alternative-specific regressors.

26.11 Count dependent-variable models

These models concern dependent variables that count the number of occurrences of an event. In
this category, we include Poisson and negative binomial regression. For the Poisson model,

E(count) = Ej exp(xj β)

where Ej is the exposure time. poisson fits this model; see [R] poisson. There is also an exact
Poisson estimator; see [U] 26.12 Exact estimators. ivpoisson fits a Poisson model where one or
more of the regressors are endogenously determined. It can also be used for modeling nonnegative
continuous outcomes instead of counts. See [R] ivpoisson
Negative binomial regression refers to estimating with data that are a mixture of Poisson counts.
One derivation of the negative binomial model is that individual units follow a Poisson regression
model but there is an omitted variable that follows a gamma distribution with parameter α. Negative
binomial regression estimates β and α. nbreg fits such models. A variation on this, unique to Stata,
allows you to model α. gnbreg fits those models. See [R] nbreg.
Truncation refers to count models in which the outcome count variable is observed only above a
certain threshold. In truncated data, the threshold is typically zero. Commands tpoisson and tnbreg
fit such models; see [R] tpoisson and [R] tnbreg.
Zero inflation refers to count models in which the number of zero counts is more than would
be expected in the regular model. The excess zeros are explained by a preliminary probit or logit
process. If the preliminary process produces a positive outcome, the usual counting process occurs,
and otherwise the count is zero. Thus whenever the preliminary process produces a negative outcome,
excess zeros are produced. The zip and zinb commands fit such models; see [R] zip and [R] zinb.

26.12 Exact estimators

Exact estimators refer to models which, rather than being estimated by asymptotic formulas, are
estimated by enumerating the conditional distribution of the sufficient statistics and then computing
the maximum likelihood estimate using that distribution. Standard errors cannot be estimated, but
confidence intervals can be and are obtained from the enumerations.
exlogistic fits logistic models of binary data in this way.
 10 [ U ] 26 Overview of Stata estimation commands

expoisson fits Poisson models of count data in this way.

In small samples, exact estimates have better coverage than the asymptotic estimates, and exact
estimates are the only way to obtain estimates, tests, and confidence intervals of covariates that
perfectly predict the observed outcome.

26.13 Linear regression with heteroskedastic errors

We now consider the model yj = xj β + j , where the variance of j is nonconstant.
First, regress can fit such models if you specify the vce(robust) option. What Stata calls
robust is also known as the White correction for heteroskedasticity.
For scientists who have data where the variance of j is known a priori, vwls is the command.
vwls produces estimates for the model given each observation’s variance, which is recorded in a
variable in the data.
If you wish to model the heteroskedasticity on covariates, use the het() option of the arch
command. Although arch is written primarily to analyze time-series data, it can be used with cross-
sectional data. Before using arch with cross-sectional data, set the data as time series, by typing gen
faketime = n and then typing tsset faketime.
Finally, qreg performs quantile regression, which in the presence of heteroskedasticity is most
of interest. Median regression (one of qreg’s capabilities) is an estimator of yj = xj β + j when
j is heteroskedastic. Even more useful, you can fit models of other quantiles and so model the
heteroskedasticity. Also see the sqreg and iqreg commands; sqreg estimates multiple quantiles
simultaneously. iqreg estimates differences in quantiles.

26.14 Stochastic frontier models

frontier fits stochastic production or cost frontier models on cross-sectional data. The model
can be expressed as

yi = xi β + vi − sui

where

1 for production functions

n
s=
−1 for cost functions
ui is a nonnegative disturbance standing for technical inefficiency in the production function or
cost inefficiency in the cost function. Although the idiosyncratic error term vi is assumed to have
a normal distribution, the inefficiency term is assumed to be one of the three distributions: half-
normal, exponential, or truncated-normal. Also, when the nonnegative component of the disturbance
is assumed to be either half-normal or exponential, frontier can fit models in which the error
components are heteroskedastic conditional on a set of covariates. When the nonnegative component
of the disturbance is assumed to be from a truncated-normal distribution, frontier can also fit a
conditional mean model, where the mean of the truncated-normal distribution is modeled as a linear
function of a set of covariates.
For panel-data stochastic frontier models, see [U] 26.18.1 Linear regression with panel data.
 [ U ] 26 Overview of Stata estimation commands 11

26.15 Regression with systems of equations

For systems of equations with endogenous covariates, use the three-stage least-squares (3SLS)
estimator reg3. The reg3 command can produce constrained and unconstrained estimates.
When we have correlated errors across equations but no endogenous right-hand-side variables,

y1j = x1j β + 1j

y2j = x2j β + 2j
..
.
ymj = xmj β + mj

where k· and l· are correlated with correlation ρkl , a quantity to be estimated from the data. This
is called Zellner’s seemingly unrelated regressions, and sureg fits such models. When x1j = x2j =
· · · = xmj , the model is known as multivariate regression, and the corresponding command is mvreg.
The equations need not be linear; if they are not linear, use nlsur.

26.16 Models with endogenous sample selection

What has become known as the Heckman model refers to linear regression in the presence of
sample selection: yj = xj β + j is not observed unless some event occurs that itself has probability
pj = F (zj γ + νj ), where  and ν might be correlated and zj and xj may contain variables in
common.
heckman fits such models by maximum likelihood or Heckman’s original two-step procedure.
This model has recently been generalized to replace the linear regression equation with another
probit equation, and that model is fit by heckprobit. heckoprobit fits an ordered probit model in
the presence of sample selection.
Another important case of endogenous sample selection is the endogenous treatment-effects model,
which considers the effect of an endogenously determined binary treatment variable on an outcome.
etregress fits an endogenous treatment-effects model by using a linear model for the outcome.
etregress uses either a two-step consistent estimator or full maximum likelihood. etpoisson fits
an endogenous treatment-effects model by using a nonlinear (exponential) model for the outcome.
etpoisson uses a likelihood estimator and can be used for modeling counts.

26.17 Models with time-series data

ARIMA refers to models with autoregressive integrated moving-average processes, and Stata’s arima
command fits models with ARIMA disturbances via the Kalman filter and maximum likelihood. These
models may be fit with or without covariates. arima also fits ARMA models. See [TS] arima.
ARFIMA stands for autoregressive fractionally integrated moving average and handles long-memory
processes. ARFIMA generalizes the ARMA and ARIMA models. ARMA models assume short memory;
after a shock, the process reverts to its trend relatively quickly. ARIMA models assume shocks are
permanent and memory never fades. ARFIMA provides a middle ground in the length of the process’s
memory. The arfima command fits ARFIMA models. In addition to one-step and dynamic forecasts,
arfima can predict fractionally integrated series. See [TS] arfima.
12 [ U ] 26 Overview of Stata estimation commands

UCM stands for unobserved components model and decomposes a time series into trend, seasonal,
cyclic, and idiosyncratic components after controlling for optional exogenous variables. UCM provides
a flexible and formal approach to smoothing and decomposition problems. The ucm command fits
UCM models. See [TS] ucm.
Relatedly, band-pass and high-pass filters are also used to decompose a time series into trend
and cyclic components, even though the tsfilter commands are not estimation commands; see
[TS] tsfilter. Provided are Baxter–King, Butterworth, Christiano–Fitzgerald, and Hodrick–Prescott
filters.
Concerning ARIMA, ARFIMA, and UCM, the estimated parameters are sometimes more easily
interpreted in terms of the implied spectral density. psdensity transforms results; see [TS] psdensity.
Stata’s prais command performs regression with AR(1) disturbances using the Prais – Winsten
or Cochrane – Orcutt transformation. Both two-step and iterative solutions are available, as well as a
version of the Hildreth – Lu search procedure. See [TS] prais.
newey produces linear regression estimates with the Newey – West variance estimates that are robust
to heteroskedasticity and autocorrelation of specified order. See [TS] newey.
Stata provides estimators for ARCH, GARCH, univariate, and multivariate models. These models
are for time-varying volatility. ARCH models allow for conditional heteroskedasticity by including
lagged variances. GARCH models also include lagged second moments of the innovations (errors).
ARCH stands for autoregressive conditional heteroskedasticity. GARCH stands for generalized ARCH.
arch fits univariate ARCH and GARCH models, and the command provides many popular extensions,
including multiplicative conditional heteroskedasticity. Errors may be normal or Student’s t or may
follow a generalized error distribution. Robust standard errors are optionally provided. See [TS] arch.
mgarch fits multivariate ARCH and GARCH models, including the diagonal vech model and the
constant, dynamic, and varying conditional correlation models. Errors may be multivariate normal or
multivariate Student’s t. Robust standard errors are optionally provided. See [TS] mgarch.
Stata provides VAR, SVAR, and VEC estimators for modeling multivariate time series. VAR and
SVAR deal with stationary series, and SVAR places additional constraints on the VAR model that
identifies the impulse–response functions. VEC is for cointegrating VAR models. VAR stands for vector
autoregression. SVAR stands for structural VAR. VEC stands for vector error-correction model.
var fits VAR models, svar fits SVAR models, and vec fits VEC models. These commands share many
of the same features for specification testing, forecasting, and parameter interpretation; see [TS] var
intro for both var and svar, [TS] vec intro for vec, and [TS] irf for all three impulse–response
functions and forecast-error variance decomposition. For lag-order selection, residual analysis, and
Granger causality tests, see [TS] var intro (for var and svar) and [TS] vec intro.
sspace estimates the parameters of multivariate state-space models using the Kalman filter. The
state-space representation of time-series models is extremely flexible and can be used to estimate
the parameters of many different models, including vector autoregressive moving-average (VARMA)
models, dynamic-factor (DF) models, and structural time-series (STS) models. It can also solve some
stochastic dynamic-programming problems. See [TS] sspace.
dfactor estimates the parameters of dynamic-factor models. These flexible models for multivariate
time series provide for a vector-autoregressive structure in both observed outcomes and in unobserved
factors. They also allow exogenous covariates for observed outcomes or unobserved factors. See
[TS] dfactor.
 [ U ] 26 Overview of Stata estimation commands 13

26.18 Panel-data models

26.18.1 Linear regression with panel data

This section could just as well be called “linear regression with complex error structures”. Commands
in this class begin with the letters xt.
xtreg fits models of the form
yit = xit β + νi + it
xtreg can produce the between-regression estimator, the within-regression (fixed effects) estimator,
or the GLS random-effects (matrix-weighted average of between and within results) estimator. It can
also produce the maximum-likelihood random-effects estimator.
xtregar can produce the within estimator and a GLS random-effects estimator when the it are
assumed to follow an AR(1) process.
xtivreg contains the between-2SLS estimator, the within-2SLS estimator, the first-differenced-2SLS
estimator, and two GLS random-effects-2SLS estimators to handle cases in which some of the covariates
are endogenous.
xtabond is for use with dynamic panel-data models (models in which there are lagged dependent
variables) and can produce the one-step, one-step robust, and two-step Arellano–Bond estimators.
xtabond can handle predetermined covariates, and it reports both the Sargan and autocorrelation tests
derived by Arellano and Bond.
xtdpdsys is an extension of xtabond and produces estimates with smaller bias when the coefficients
of the AR process are large. xtpdsys is also more efficient than xtabond. Whereas xtabond uses
moment conditions based on the differenced errors, xtpdsys uses moment conditions based on both
the differenced errors and their levels.
xtdpd is an extension of xtdpdsys and can be used to estimate the parameters of a broader class
of dynamic panel-data models. xtdpd can be used to fit models with serially correlated idiosyncratic
errors, whereas xtdpdsys and xtabond assume no serial correlation. Or xtdpd can be used with
models where the structure of the predetermined variables is more complicated than that assumed by
xtdpdsys or xtabond.
xtgls produces generalized least-squares estimates for models of the form

yit = xit β + it

where you may specify the variance structure of it . If you specify that it is independent for all i and
t, xtgls produces the same results as regress up to a small-sample degrees-of-freedom correction
applied by regress but not by xtgls.
You may choose among three variance structures concerning i and three concerning t, producing
a total of nine different models. Assumptions concerning i deal with heteroskedasticity and cross-
sectional correlation. Assumptions concerning t deal with autocorrelation and, more specifically, AR(1)
serial correlation.
Alternative methods report the OLS coefficients and a version of the GLS variance–covariance
estimator. xtpcse produces panel-corrected standard error (PCSE) estimates for linear cross-sectional
time-series models, where the parameters are estimated by OLS or Prais–Winsten regression. When
you are computing the standard errors and the variance–covariance estimates, the disturbances are,
by default, assumed to be heteroskedastic and contemporaneously correlated across panels.
 14 [ U ] 26 Overview of Stata estimation commands

In the jargon of GLS, the random-effects model fit by xtreg has exchangeable correlation
within i — xtgls does not model this particular correlation structure. xtgee, however, does.
xtgee fits population-averaged models, and it optionally provides robust estimates of variance.
Moreover, xtgee allows other correlation structures. One that is of particular interest to those with
many data goes by the name unstructured. The within-panel correlations are simply estimated in an
unconstrained way. [U] 26.18.3 Generalized linear models with panel data will discuss this estimator
further because it is not restricted to linear regression models.
xthtaylor uses instrumental variables estimators to estimate the parameters of panel-data random-
effects models of the form

yit = X1it β1 + X2it β2 + Z1i δ1 + Z2i δ2 + ui + eit

The individual effects ui are correlated with the explanatory variables X2it and Z2i but are uncorrelated
with X1it and Z1i , where Z1 and Z2 are constant within panel.
xtfrontier fits stochastic production or cost frontier models for panel data. You may choose from
a time-invariant model or a time-varying decay model. In both models, the nonnegative inefficiency
term is assumed to have a truncated-normal distribution. In the time-invariant model, the inefficiency
term is constant within panels. In the time-varying decay model, the inefficiency term is modeled as
a truncated-normal random variable multiplied by a specific function of time. In both models, the
idiosyncratic error term is assumed to have a normal distribution. The only panel-specific effect is
the random inefficiency term.
See [U] 26.19 Multilevel mixed-effects models for a generalization of xtreg that allows for
multiple levels of panels, random coefficients, and variance-component estimation in general.

26.18.2 Censored linear regression with panel data

xttobit fits random-effects tobit models and generalizes that to observation-specific censoring.
xtintreg performs random-effects interval regression and generalizes that to observation-specific
censoring. Interval regression, in addition to allowing open-ended intervals, also allows closed intervals.

26.18.3 Generalized linear models with panel data

[U] 26.6 Generalized linear models above discussed the model

g{E(yj )} = xj β, yj ∼ F (1)

where g() is the link function and F is a member of the exponential family, both of which you
specify before estimation.
There are two ways to extend the generalized linear model to panel data. They are the generalized
linear mixed model (GLMM) and generalized estimation equations (GEE).
GEE uses a working correlation structure to model within-panel correlation. GEEs may be fit with
the xtgee command; see [XT] xtgee.
For generalized linear models with multilevel data, including panel data, see [U] 26.19 Multilevel
mixed-effects models.
 [ U ] 26 Overview of Stata estimation commands 15

26.18.4 Qualitative dependent-variable models with panel data

xtprobit fits random-effects probit regression via maximum likelihood. It also fits population-
averaged models via GEE. This last is nothing more than xtgee with the binomial family, probit link,
and exchangeable error structure.
xtlogit fits random-effects logistic regression models via maximum likelihood. It also fits
conditional fixed-effects models via maximum likelihood. Finally, as with xtprobit, it fits population-
averaged models via GEE.
xtcloglog estimates random-effects complementary log-log regression via maximum likelihood.
It also fits population-averaged models via GEE.
xtologit and xtoprobit are multiple-outcome models. xtologit fits a random-effects ordered
logistic model, and xtoprobit fits a random-effects ordered probit model.
These models are generalizable to multilevel data; see [U] 26.19 Multilevel mixed-effects models.

26.18.5 Count dependent-variable models with panel data

xtpoisson fits two different random-effects Poisson regression models via maximum likelihood.
The two distributions for the random effect are gamma and normal. It also fits conditional fixed-effects
models, and it fits population-averaged models via GEE. This last is nothing more than xtgee with
the Poisson family, log link, and exchangeable error structure.
xtnbreg fits random-effects negative binomial regression models via maximum likelihood (the
distribution of the random effects is assumed to be beta). It also fits conditional fixed-effects models,
and it fits population-averaged models via GEE.
These models are generalizable to multilevel data; see [U] 26.19 Multilevel mixed-effects models.

26.18.6 Random-coefficients model with panel data

xtrc fits Swamy’s random-coefficients linear regression model. In this model, rather than only
the intercept varying across groups, all the coefficients are allowed to vary. xtrc is a special case of
mixed.

26.19 Multilevel mixed-effects models

In multilevel data, observations—subjects, for want of a better word—can be divided into groups
that have something in common. Perhaps the subjects are students, and the groups attended the same
high school, or they are patients who were treated at the same hospital, or they are tractors that
were manufactured at the same factory. Whatever it is they have in common, it may be reasonable
to assume that the shared attribute affects the outcome being modeled.
With regard to students and high school, perhaps you are modeling later success in life. Some
high schools are better (or worse) than others, so it would not be unreasonable to assume that the
identity of the high school had an effect. With regard to patients and hospital, the argument is much
the same if the outcome is subsequent health: some hospitals are better (or worse) than others, at
least with respect to particular health problems. With regard to tractors and factory, it would hardly
be surprising if tractors from some factories were more reliable than tractors from other factories.
Described above is two-level data. The first level is the student, patient, or tractor, and the second
level is the high school, hospital, or factory. Observations are said to be nested within groups: students
within a high school, patients within a hospital, or tractors within a factory.
16 [ U ] 26 Overview of Stata estimation commands

Even though the effect on outcome is not directly observed, one can control for the effect if one
is willing to assume that the effect is the same for all observations within a group and that, across
groups, the effect is a random draw from a statistical distribution that is uncorrelated with the overall
residual of the model and other group effects.
We have just described multilevel models.
A more complicated scenario might have three levels: students nested within teachers within a
high school, patients nested within doctors within a hospital, or tractors nested within an assembly
line within a factory.
An alternative to three-level hierarchical data is crossed data. We have workers and their occupation
and the industry in which they work.
In any case, multilevel data arise in a variety of situations. One possible way of analyzing such
data is simply to ignore the multilevel aspect of the data. If you do that, and assuming that the ignored
effect is uncorrelated with the residual, you will still obtain unbiased coefficients, although standard
errors produced by standard methods will be incorrect. Many estimation commands in Stata provide
cluster–robust standard errors to get around that problem.
You can obtain more efficient parameter estimates, however, if you use an estimator that explicitly
accounts for the multilevel nature of the data. And if you want to perform comparisons across groups,
you must use such estimators.
Stata provides a suite of multilevel estimation commands. The estimation commands are the
following:

Command Outcome variable Equivalent to

mixed continuous linear regression
meprobit binary probit regression
melogit binary logistic regression
meqrlogit binary logistic regression 1
mecloglog binary complementary log-log regression
meoprobit ordered categorical ordered probit regression
meologit ordered categorical ordered logistic regression
mepoisson count Poisson regression
meqrpoisson count Poisson regression 1
menbreg count negative binomial regression
meglm various generalized linear models
1
meqrlogit and meqrpoisson use QR decomposition to produce results. They exist for
historical reasons but can be useful for estimation on the boundary of the parameter space.
Results are numerically equivalent to melogit and mepoisson in other cases.

The above estimators provide random intercepts and random coefficients and allow constraints
to be placed on coefficients and on variance components. (The QR decomposition estimators do not
allow constraints.)
See the [ME] Stata Multilevel Mixed-Effects Reference Manual; in particular, see [ME] me.
 [ U ] 26 Overview of Stata estimation commands 17

26.20 Survival-time (failure-time) models

Commands are provided to fit Cox proportional-hazards models, competing-risks regression, and
several parametric survival models including exponential, Weibull, Gompertz, lognormal, loglogistic,
and generalized gamma; see [ST] stcox, [ST] stcrreg, and [ST] streg. The commands for Cox and
parametric regressions, stcox and streg, are appropriate for single- or multiple-failure-per-subject
data. The command for competing-risks regression, stcrreg, is appropriate only for single-failure data.
Conventional, robust, bootstrap, and jackknife standard errors are available with all three commands,
with the exception that for stcrreg, robust standard errors are the conventional standard errors.
Both the Cox model and the parametric models (as fit using Stata) allow for two additional
generalizations. First, the models may be modified to allow for latent random effects, or frailties.
Second, the models may be stratified in that the baseline hazard function may vary completely over
a set of strata. The parametric models also allow for the modeling of ancillary parameters.
Competing-risks regression, as fit using Stata, is a useful alternative to Cox regression for datasets
where more than one type of failure occurs, in other words, for data where failure events compete
with one another. In such situations, competing-risks regression allows you to easily assess covariate
effects on the incidence of the failure type of interest without having to make strong assumptions
concerning the independence of failure types.
stcox, stcrreg, and streg require that the data be stset so that the proper response variables
may be established. After you stset the data, the time/censoring response is taken as understood,
and you need only supply the regressors (and other options) to stcox, stcrreg, and streg. With
stcrreg, one required option deals with specifying which events compete with the failure event of
interest that was previously stset.

26.21 Treatment-effect models

teffects estimates treatment effects from observational data.
A treatment effect is the change in an outcome caused by an individual getting one treatment
instead of another. We can estimate average treatment effects, but not individual-level treatment effects,
because we only observe each individual getting one or another treatment.
teffects uses methods that specify what the individual-level outcomes would be for each treatment
level, even though only one of them can be realized. This approach is known as the potential-outcome
framework.
It is not appropriate to use teffects when a treatment is endogenously determined (the potential
outcomes are not conditionally independent). When the treatment is endogenous, an endogenous
treatment-effects model can be used to estimate the average treatment effect. These models consider
the effect of an endogenously determined binary treatment variable on the outcome. etregress fits
an endogenous treatment-effects model by using a linear model for the outcome. etpoisson fits an
endogenous treatment-effects model by using nonlinear (exponential) model for the outcome.
Suppose we want to use observational data to learn about the effect of exercise on blood pressure.
The potential-outcome framework provides the structure to estimate what would be the average effect
of everyone exercising instead of everyone not exercising, an effect known as average treatment effect
(ATE). Similarly, we can estimate the average effect, among those who exercise, of exercising instead
of not exercising, which is known as the average treatment effect on the treated (ATET). Finally, we
could estimate the average blood pressure that would be obtained if everyone exercised or if no one
exercised, parameters known as potential-outcome means (POMs).
teffects can estimate the ATE, the ATET, and the POMs. The estimators implemented in teffects
impose the structure of the potential-outcome framework on the data in different ways.
 18 [ U ] 26 Overview of Stata estimation commands

• Regression-adjustment estimators use models for the potential outcomes.

• Inverse-probability-weighted estimators use models for treatment assignment.
• Augmented inverse-probability-weighted estimators and inverse-probability-weighted regression-
adjustment estimators use models for the potential outcomes and for treatment assignment. These
estimators have the double-robust property; they correctly estimate the treatment effect even if
only one of the two models is correctly specified.
• Nearest-neighbor matching (NNM) and propensity-score matching (PSM) estimators compare the
outcomes of individuals who are similar as possible except that one gets the treatment and the
other does not. NNM uses a nonparametric similarity measure, while PSM uses estimated treatment
probabilities to measure similarity.
etregress and etpoisson can be used to estimate the ATE and the ATET. etregress uses either
a two-step consistent estimator or full maximum likelihood. etpoisson uses a maximum likelihood
estimator. See [TE] etregress and [TE] etpoisson.
See [TE] teffects intro or [TE] teffects intro advanced for a simple or an advanced introduction.
teffects can estimate treatment effects from multivalued treatments; see [TE] teffects multivalued.
For examples and details, see [TE] teffects aipw, [TE] teffects ipw, [TE] teffects ipwra, [TE] teffects
nnmatch, [TE] teffects psmatch, or [TE] teffects ra.

26.22 Generalized method of moments (GMM)

gmm fits models using generalized method of moments (GMM). With the interactive version of the
command, you enter your moment equations directly into the dialog box or command line using
substitutable expressions just like with nl or nlsur. The moment-evaluator program version gives
you greater flexibility in exchange for increased complexity; with this version, you write a program
that calculates the moments based on a vector of parameters passed to it.
gmm can fit both single- and multiple-equation models, and you can combine moment conditions of
the form E{zi ui (β)} = 0, where zi is a vector of instruments and ui (β) is often an additive regression
error term, as well as more general moment conditions of the form E{hi (zi ; β)} = 0. In the former
case, you specify the expression for ui (β) and use the instruments() and xtinstruments()
options to specify zi . In the latter case, you specify the expression for hi (zi ; β); because that
expression incorporates your instruments, you do not use the instruments() or xtinstruments()
option.
gmm supports cross-sectional, time-series, and panel data. You can request weight matrices and VCEs
that are suitable for independent and identically distributed errors, that are suitable for heteroskedastic
errors, that are appropriate for clustered observations, or that are heteroskedasticity- and autocorrelation-
consistent (HAC). For HAC weight matrices and VCEs, gmm lets you specify the bandwidth or request
an automatic bandwidth selection algorithm.

26.23 Estimation with correlated errors

By correlated errors, we mean that observations are grouped, and that within group, the observations
might be correlated but, across groups, they are uncorrelated. regress with the vce(cluster
clustvar) option can produce “correct” estimates, that is, inefficient estimates with correct standard
errors and lots of robustness; see [U] 20.21 Obtaining robust variance estimates. Obviously, if you
know the correlation structure (and are not mistaken), you can do better, so xtreg and xtgls are
also of interest here; we discuss them in [U] 26.18.1 Linear regression with panel data.
Estimation in the presence of autocorrelated errors is discussed in [U] 26.17 Models with time-series
data.
 [ U ] 26 Overview of Stata estimation commands 19

26.24 Survey data

Stata’s svy command fits statistical models for complex survey data. svy is a prefix command,
so to obtain linear regression, you type
. svy: regress . . .
or to obtain probit regression, you type
. svy: probit . . .
but first you must type a svyset command to define the survey design characteristics. Prefix svy
works with many estimation commands, and everything is documented together in the Stata Survey
Data Reference Manual.
svy supports the following variance-estimation methods:
• Taylor-series linearization
• Bootstrap
• Balanced repeated replication (BRR)
• Jackknife
• Successive difference replication (SDR)
See [SVY] variance estimation for details.
svy supports the following survey design characteristics:
• With- and without- replacement sampling
• Sampling weights
• Stratification
• Poststratification
• Clustering
• Multiple stages of clustering without replacement
• BRR and jackknife replication weights
See [SVY] svyset for details.
Subpopulation estimation is available for all estimation commands.
Tabulations and summary statistics are also available, including means, proportions, ratios, and
totals over multiple subpopulations, and direct standardization of means, proportions, and ratios.
See [SVY] survey.

26.25 Multiple imputation

Multiple imputation (MI) is a statistical technique for estimation in the presence of missing data.
If you fit the parameters of y on x1 , x2 , and x3 using any of the other Stata estimation commands,
parameters are fit on the data for which y , x1 , x2 , and x3 contain no missing values. This process
is known as listwise or casewise deletion because observations for which any of y , x1 , x2 , or x3
contain missing values are ignored or, said differently, deleted from consideration. MI is a technique to
recover the information in those ignored observations when the missing values are missing at random
(MAR) or missing completely at random (MCAR). Data are MAR if the probability that a value is
missing may depend on observed data but not on unobserved data. Data are MCAR if the probability
of missingness is not even a function of the observed data.
 20 [ U ] 26 Overview of Stata estimation commands

MI is named for the imputations it produces to replace the missing values in the data. MI does not
just form replacement values for the missing data, it produces multiple replacements. The purpose is
not to create replacement values as close as possible to the true ones, but to handle missing data in
a way resulting in valid statistical inference.
There are three steps in an MI analysis. First, one forms M imputations for each missing value
in the data. Second, one fits the model of interest separately on each of the M resulting datasets.
Finally, one combines those M estimation results into the desired single result.
The mi command does this for you. It can be used with most of Stata’s estimation commands,
including survey, survival, and panel and multilevel models. See [MI] intro.

26.26 Multivariate and cluster analysis

Most of Stata’s multivariate capabilities are to be found in the Multivariate Statistics Reference Manual,
although there are some exceptions.
1. mvreg fits multivariate regressions.
2. manova fits MANOVA and MANCOVA models, one-way and up—including two-way factorial, three-
way factorial, etc.—and it fits nested and mixed-design models. Also see [U] 26.5 ANOVA,
ANCOVA, MANOVA, and MANCOVA above.
3. canon estimates canonical correlations and their corresponding loadings. Canonical correlation
attempts to describe the relationship between two sets of variables.
4. pca extracts principal components and reports eigenvalues and loadings. Some people consider
principal components a descriptive tool—in which case standard errors as well as coefficients are
relevant—and others look at it as a dimension-reduction technique.
5. factor fits factor models and provides principal factors, principal-component factors, iterated
principal-component factors, and maximum-likelihood solutions. Factor analysis is concerned with
finding few common factors b zk , k = 1, . . . , q that linearly reconstruct the original variables yi ,
i = 1, . . . , L.
6. tetrachoric, in conjunction with pca or factor, allows you to perform PCA or factor analysis
on binary data.
7. rotate provides a wide variety of orthogonal and oblique rotations after factor and pca.
Rotations are often used to produce more interpretable results.
8. procrustes performs Procrustes analysis, one of the standard methods of multidimensional scaling.
It can perform orthogonal or oblique rotations, as well as translation and dilation.
9. mds performs metric and nonmetric multidimensional scaling for dissimilarity between observations
with respect to a set of variables. A wide variety of dissimilarity measures are available and, in
fact, are the same as those for cluster.
10. ca performs correspondence analysis, an exploratory multivariate technique for analyzing cross-
tabulations and the relationship between rows and columns.
11. mca performs multiple correspondence analysis (MCA) and joint correspondence analysis (JCA).
12. mvtest performs tests of multivariate normality along with tests of means, covariances, and
correlations.
13. cluster provides cluster analysis; both hierarchical and partition clustering methods are available.
Strictly speaking, cluster analysis does not fall into the category of statistical estimation. Rather, it
is a set of techniques for exploratory data analysis. Stata’s cluster environment has many different
similarity and dissimilarity measures for continuous and binary data.
 [ U ] 26 Overview of Stata estimation commands 21

14. discrim and candisc perform discriminant analysis. candisc performs linear discriminant
analysis (LDA). discrim also performs LDA, and it performs quadratic discriminant analysis
(QDA), k th nearest neighbor (KNN), and logistic discriminant analysis. The two commands differ
in default output. discrim shows the classification summary, candisc shows the canonical linear
discriminant functions, and both will produce either.

26.27 Pharmacokinetic data

There are four estimation commands for analyzing pharmacokinetic data. See [R] pk for an overview
of the pk system.
1. pkexamine calculates pharmacokinetic measures from time-and-concentration subject-level data.
pkexamine computes and displays the maximum measured concentration, the time at the maximum
measured concentration, the time of the last measurement, the elimination time, the half-life, and
the area under the concentration–time curve (AUC).
2. pksumm obtains the first four moments from the empirical distribution of each pharmacokinetic
measurement and tests the null hypothesis that the distribution of that measurement is normally
distributed.
3. pkcross analyzes data from a crossover design experiment. When one is analyzing pharmaceutical
trial data, if the treatment, carryover, and sequence variables are known, the omnibus test for
separability of the treatment and carryover effects is calculated.
4. pkequiv performs bioequivalence testing for two treatments. By default, pkequiv calculates a
standard confidence interval symmetric about the difference between the two treatment means.
pkequiv also calculates confidence intervals symmetric about zero and intervals based on Fieller’s
theorem. Also, pkequiv can perform interval hypothesis tests for bioequivalence.

26.28 Specification search tools

There are three other commands that are not really estimation commands but are combined with
estimation commands to assist in specification searches: stepwise, fp, and mfp.
stepwise, one of Stata’s prefix commands, provides stepwise estimation. You can use the stepwise
prefix with some, but not all, estimation commands. See [R] stepwise for a list of estimation commands
that are supported.
fp and mfp are commands to assist you in performing fractional-polynomial functional specification
searches.

26.29 Power and sample-size analysis

Power and sample-size (PSS) analysis is performed during the planning stage of a study. The main
goal of PSS analysis is to provide an estimate of the sample size needed to successfully achieve the
research objective of a study.
For example, suppose that we want to design a study to evaluate a new drug for lowering blood
pressure. We want to test whether the mean blood pressure of the experimental group, which will
receive the new drug, is the same as the mean blood pressure of the control group, which will receive
the old drug. The post-hoc analysis will use a two-sample t test to test the difference between the
two means. How many subjects do we need to enroll in our study to detect a difference between
means that is of clinical importance? PSS analysis can answer this question.
 22 [ U ] 26 Overview of Stata estimation commands

PSS analysis can also answer other questions that may arise during the planning stage of a study.
For example, what is the power of a test given an available sample size, and how likely is it to
detect an effect of interest given limited study resources? The answers to these questions may help
reduce the cost of a study by preventing an overpowered study or may avoid wasting resources on
an underpowered study.
See [PSS] intro for more information about PSS analysis.
The power command performs PSS analysis. It provides PSS analysis for comparison of means,
variances, proportions, and correlations. One-sample, two-sample, and paired analyses are supported.
power provides both tabular output and graphical output, or power curves; see [PSS] power, table
and [PSS] power, graph for details.
See [PSS] power for a full list of supported methods and the description of the command.
You can work with power commands either interactively or via a convenient point-and-click
interface; see [PSS] GUI for details.

26.30 Obtaining new estimation commands

This chapter has discussed all the official estimation commands included in Stata 13. Users may
have written their own estimation commands that they are willing to share. Type search estimation,
ssc new, and ssc hot to discover more estimation commands; see [R] ssc.
And, of course, you can always write your own commands; see [R] ml.

26.31 References
Gould, W. W. 2000. sg124: Interpreting logistic regression in all its forms. Stata Technical Bulletin 53: 19–29.
Reprinted in Stata Technical Bulletin Reprints, vol. 9, pp. 257–270. College Station, TX: Stata Press.
. 2011. Use poisson rather than regress; tell a friend. The Stata Blog: Not Elsewhere Classified.
http://blog.stata.com/2011/08/22/use-poisson-rather-than-regress-tell-a-friend/.